Zero-Shot Learning via Discriminative Dual Semantic Auto-Encoder
Zero-shot learning (ZSL) is an effective method to perform the recognition task without any training samples of specific classes. Most existing ZSL models put emphasis on learning an embedding between visual space and semantic space directly. However, few ZSL models research whether the human-design...
| Main Authors: | , , , , |
|---|---|
| Format: | Article |
| Language: | English |
| Published: |
IEEE
2021-01-01
|
| Series: | IEEE Access |
| Subjects: | |
| Online Access: | https://ieeexplore.ieee.org/document/9303397/ |
| _version_ | 1798002004726906880 |
|---|---|
| author | Nan Xing Yang Liu Hong Zhu Jing Wang Jungong Han |
| author_facet | Nan Xing Yang Liu Hong Zhu Jing Wang Jungong Han |
| author_sort | Nan Xing |
| collection | DOAJ |
| description | Zero-shot learning (ZSL) is an effective method to perform the recognition task without any training samples of specific classes. Most existing ZSL models put emphasis on learning an embedding between visual space and semantic space directly. However, few ZSL models research whether the human-designed semantic features are discriminative enough to recognize different classes. Moreover, one-way mapping suffers from the project domain shift problem. In this article, we propose to learn a Discriminative Dual Semantic Auto-encoder (DDSA) based on the encoder-decoder paradigm to solve this problem. DDSA attempts to construct two bidirectional embeddings to connect the visual space and the semantic space with the help of the learned aligned space which includes discriminative information of the visual features and semantic features. Based on the DDSA, we additionally propose a Deep DDSA to capture deep aligned features that are more conducive to zero-shot classification. The key to the proposed framework is that it implicitly exact the principal information from visual space and semantic space to construct aligned features, which is not only semantic-preserving but also discriminative. Extensive experiments on five benchmarks (SUN, CUB, AWA1, AWA2 and aPY) demonstrate the effectiveness of the proposed framework with state-of-the-art performance obtained on both conventional ZSL and generalized ZSL settings. |
| first_indexed | 2024-04-11T11:45:17Z |
| format | Article |
| id | doaj.art-41f4034b1a904e8390f9a4a7dd7348d0 |
| institution | Directory Open Access Journal |
| issn | 2169-3536 |
| language | English |
| last_indexed | 2024-04-11T11:45:17Z |
| publishDate | 2021-01-01 |
| publisher | IEEE |
| record_format | Article |
| series | IEEE Access |
| spelling | doaj.art-41f4034b1a904e8390f9a4a7dd7348d02022-12-22T04:25:37ZengIEEEIEEE Access2169-35362021-01-01973374210.1109/ACCESS.2020.30465739303397Zero-Shot Learning via Discriminative Dual Semantic Auto-EncoderNan Xing0Yang Liu1https://orcid.org/0000-0001-5917-8653Hong Zhu2https://orcid.org/0000-0003-2993-1928Jing Wang3Jungong Han4https://orcid.org/0000-0003-4361-956XSchool of Automation and Information Engineering, Xi’an University of Technology, Xi’an, ChinaState Key Laboratory of Integrated Services Networks, Xidian University, Xi’an, ChinaSchool of Automation and Information Engineering, Xi’an University of Technology, Xi’an, ChinaFaculty of Printing Packaging Engineering and Digital Media Technology, Xi’an University of Technology, Xi’an, ChinaComputer Science Department, Aberystwyth University, Aberystwyth, U.K.Zero-shot learning (ZSL) is an effective method to perform the recognition task without any training samples of specific classes. Most existing ZSL models put emphasis on learning an embedding between visual space and semantic space directly. However, few ZSL models research whether the human-designed semantic features are discriminative enough to recognize different classes. Moreover, one-way mapping suffers from the project domain shift problem. In this article, we propose to learn a Discriminative Dual Semantic Auto-encoder (DDSA) based on the encoder-decoder paradigm to solve this problem. DDSA attempts to construct two bidirectional embeddings to connect the visual space and the semantic space with the help of the learned aligned space which includes discriminative information of the visual features and semantic features. Based on the DDSA, we additionally propose a Deep DDSA to capture deep aligned features that are more conducive to zero-shot classification. The key to the proposed framework is that it implicitly exact the principal information from visual space and semantic space to construct aligned features, which is not only semantic-preserving but also discriminative. Extensive experiments on five benchmarks (SUN, CUB, AWA1, AWA2 and aPY) demonstrate the effectiveness of the proposed framework with state-of-the-art performance obtained on both conventional ZSL and generalized ZSL settings.https://ieeexplore.ieee.org/document/9303397/Zero-shot learningdiscriminativeencoder-decoderaligned |
| spellingShingle | Nan Xing Yang Liu Hong Zhu Jing Wang Jungong Han Zero-Shot Learning via Discriminative Dual Semantic Auto-Encoder IEEE Access Zero-shot learning discriminative encoder-decoder aligned |
| title | Zero-Shot Learning via Discriminative Dual Semantic Auto-Encoder |
| title_full | Zero-Shot Learning via Discriminative Dual Semantic Auto-Encoder |
| title_fullStr | Zero-Shot Learning via Discriminative Dual Semantic Auto-Encoder |
| title_full_unstemmed | Zero-Shot Learning via Discriminative Dual Semantic Auto-Encoder |
| title_short | Zero-Shot Learning via Discriminative Dual Semantic Auto-Encoder |
| title_sort | zero shot learning via discriminative dual semantic auto encoder |
| topic | Zero-shot learning discriminative encoder-decoder aligned |
| url | https://ieeexplore.ieee.org/document/9303397/ |
| work_keys_str_mv | AT nanxing zeroshotlearningviadiscriminativedualsemanticautoencoder AT yangliu zeroshotlearningviadiscriminativedualsemanticautoencoder AT hongzhu zeroshotlearningviadiscriminativedualsemanticautoencoder AT jingwang zeroshotlearningviadiscriminativedualsemanticautoencoder AT jungonghan zeroshotlearningviadiscriminativedualsemanticautoencoder |